Modify feature on execute.

ge/graph/load/new_model_manager/davinci_model.cc

@@ -108,6 +108,7 @@ std::mutex DavinciModel::tvm_bin_mutex_;
 DavinciModel::DavinciModel(int32_t priority, const std::shared_ptr<ModelListener> &listener)
     : weights_mem_base_(nullptr),
       var_mem_base_(nullptr),
+      fixed_mem_base_(0),
       mem_base_(nullptr),
       is_inner_mem_base_(false),
       is_inner_weight_base_(false),
@@ -670,6 +671,7 @@ Status DavinciModel::Init(void *dev_ptr, size_t mem_size, void *weight_ptr, size
     data_inputer_ = new (std::nothrow) DataInputer();
     GE_CHK_BOOL_RET_STATUS(data_inputer_ != nullptr, MEMALLOC_FAILED, "data_inputer_ is nullptr.");
   }
+  fixed_mem_base_ = reinterpret_cast<uintptr_t>(mem_base_);
   GE_TIMESTAMP_END(InitModelMem, "GraphLoader::InitModelMem");
 
   for (const ge::NodePtr &node : compute_graph->GetDirectNode()) {
@@ -2828,7 +2830,32 @@ Status DavinciModel::CreateKnownZeroCopyMap(const vector<void *> &inputs, const
   return SUCCESS;
 }
 
+void DavinciModel::SetTotalIOAddrs(const vector<void *> &io_addrs) {
+  if (fixed_mem_base_ == reinterpret_cast<uintptr_t>(mem_base_)) {
+    total_io_addrs_.insert(total_io_addrs_.end(), io_addrs.begin(), io_addrs.end());
+    return;
+  }
+
+  for (size_t i = 0; i < io_addrs.size(); ++i) {
+    uintptr_t addr = reinterpret_cast<uintptr_t>(io_addrs[i]);
+    if ((fixed_mem_base_ <= addr) && (addr < fixed_mem_base_ + runtime_param_.mem_size)) {
+      total_io_addrs_.emplace_back(mem_base_ + (addr - fixed_mem_base_));
+    } else {
+      total_io_addrs_.emplace_back(io_addrs[i]);
+    }
+  }
+}
+
 Status DavinciModel::UpdateKnownZeroCopyAddr(vector<void *> &total_io_addrs) {
+  if (fixed_mem_base_ != reinterpret_cast<uintptr_t>(mem_base_)) {
+    for (size_t i = 0; i < total_io_addrs.size(); ++i) {
+      uintptr_t addr = reinterpret_cast<uintptr_t>(total_io_addrs[i]);
+      if ((fixed_mem_base_ <= addr) && (addr < fixed_mem_base_ + runtime_param_.mem_size)) {
+        total_io_addrs[i] = mem_base_ + (addr - fixed_mem_base_);
+      }
+    }
+  }
+
   for (size_t i = 0; i < total_io_addrs.size(); ++i) {
     auto it_in = knonw_input_data_info_.find(total_io_addrs[i]);
     if (it_in != knonw_input_data_info_.end()) {

ge/graph/load/new_model_manager/davinci_model.h

@@ -503,9 +503,7 @@ class DavinciModel {
     void *cur_args = static_cast<char *>(args_) + offset;
     return cur_args;
   }
-  void SetTotalIOAddrs(const vector<void *> &io_addrs) {
-    total_io_addrs_.insert(total_io_addrs_.end(), io_addrs.begin(), io_addrs.end());
-  }
+  void SetTotalIOAddrs(const vector<void *> &io_addrs);
   void SetHybridArgsSize(uint32_t args_size) { total_hybrid_args_size_ += args_size; }
   uint32_t GetHybridArgsSize() {
     return total_hybrid_args_size_;
@@ -555,6 +553,7 @@ class DavinciModel {
   uint8_t *weights_mem_base_;
   uint8_t *var_mem_base_;
   // memory address of model
+  uintptr_t fixed_mem_base_;  // Initial of mem_base_, keep forever.
   uint8_t *mem_base_;
   uint8_t *p2p_mem_base_;
   bool is_inner_mem_base_;